Hugging Face's logo

qualcomm
/
MiniLM-v2

Text Generation
PyTorch
foundation
real_time
android
Model card Files
xet
 Community
qaihm-bot commited on
Commit
7218b03
·
verified ·
1 Parent(s): b049399

v0.57.0

Browse files

See https://github.com/qualcomm/ai-hub-models/releases/v0.57.0 for changelog.

Files changed (3) hide show
  1. LICENSE +1 -0
  2. README.md +168 -0
  3. release_assets.json +53 -0
LICENSE ADDED
@@ -0,0 +1 @@
 
 
1
+ The license of the original trained model can be found at https://github.com/UKPLab/sentence-transformers/blob/master/LICENSE.
README.md ADDED
@@ -0,0 +1,168 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ library_name: pytorch
3
+ license: other
4
+ tags:
5
+ - foundation
6
+ - real_time
7
+ - android
8
+ pipeline_tag: text-generation
9
+
10
+ ---
11
+
12
+ ![](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/minilm_v2/web-assets/model_demo.png)
13
+
14
+ # MiniLM-v2: Optimized for Qualcomm Devices
15
+
16
+ All-MiniLM-L6-v2 maps sentences to a 384-dimensional dense vector space. Trained on 1B+ sentence pairs, it excels at semantic search, clustering, and sentence similarity tasks while being small enough to run on mobile devices.
17
+
18
+ This is based on the implementation of MiniLM-v2 found [here](https://github.com/UKPLab/sentence-transformers).
19
+ This repository contains pre-exported model files optimized for Qualcomm® devices. You can use the [Qualcomm® AI Hub Models](https://github.com/qualcomm/ai-hub-models/blob/v0.57.0/src/qai_hub_models/models/minilm_v2) library to export with custom configurations. More details on model performance across various devices, can be found [here](#performance-summary).
20
+
21
+ Qualcomm AI Hub Models uses [Qualcomm AI Hub Workbench](https://workbench.aihub.qualcomm.com) to compile, profile, and evaluate this model. [Sign up](https://myaccount.qualcomm.com/signup) to run these models on a hosted Qualcomm® device.
22
+
23
+ ## Getting Started
24
+ There are two ways to deploy this model on your device:
25
+
26
+ ### Option 1: Download Pre-Exported Models
27
+
28
+ Below are pre-exported model assets ready for deployment.
29
+
30
+ | Runtime | Precision | Chipset | SDK Versions | Download |
31
+ |---|---|---|---|---|
32
+ | ONNX | float | Universal | QAIRT 2.45, ONNX Runtime 1.25.0 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/minilm_v2/releases/v0.57.0/minilm_v2-onnx-float.zip)
33
+ | ONNX | w8a8 | Universal | QAIRT 2.45, ONNX Runtime 1.25.0 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/minilm_v2/releases/v0.57.0/minilm_v2-onnx-w8a8.zip)
34
+ | QNN_DLC | float | Universal | QAIRT 2.45 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/minilm_v2/releases/v0.57.0/minilm_v2-qnn_dlc-float.zip)
35
+ | QNN_DLC | w8a8 | Universal | QAIRT 2.45 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/minilm_v2/releases/v0.57.0/minilm_v2-qnn_dlc-w8a8.zip)
36
+ | TFLITE | float | Universal | QAIRT 2.45 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/minilm_v2/releases/v0.57.0/minilm_v2-tflite-float.zip)
37
+ | TFLITE | w8a8 | Universal | QAIRT 2.45 | [Download](https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/minilm_v2/releases/v0.57.0/minilm_v2-tflite-w8a8.zip)
38
+
39
+ For more device-specific assets and performance metrics, visit **[MiniLM-v2 on Qualcomm® AI Hub](https://aihub.qualcomm.com/models/minilm_v2)**.
40
+
41
+
42
+ ### Option 2: Export with Custom Configurations
43
+
44
+ Use the [Qualcomm® AI Hub Models](https://github.com/qualcomm/ai-hub-models/blob/v0.57.0/src/qai_hub_models/models/minilm_v2) Python library to compile and export the model with your own:
45
+ - Custom weights (e.g., fine-tuned checkpoints)
46
+ - Custom input shapes
47
+ - Target device and runtime configurations
48
+
49
+ This option is ideal if you need to customize the model beyond the default configuration provided here.
50
+
51
+ See our repository for [MiniLM-v2 on GitHub](https://github.com/qualcomm/ai-hub-models/blob/v0.57.0/src/qai_hub_models/models/minilm_v2) for usage instructions.
52
+
53
+ ## Model Details
54
+
55
+ **Model Type:** Model_use_case.text_generation
56
+
57
+ **Model Stats:**
58
+ - Model checkpoint: sentence-transformers/all-MiniLM-L6-v2
59
+ - Input resolution: 128 tokens
60
+ - Number of parameters: 22.7M
61
+ - Model size (float): 86.7 MB
62
+ - Embedding dimension: 384
63
+
64
+ ## Performance Summary
65
+ | Model | Runtime | Precision | Chipset | Inference Time (ms) | Peak Memory Range (MB) | Primary Compute Unit
66
+ |---|---|---|---|---|---|---
67
+ | MiniLM-v2 | ONNX | float | Snapdragon® X2 Elite | 0.774 ms | 212 - 212 MB | NPU
68
+ | MiniLM-v2 | ONNX | float | Snapdragon® X Elite | 1.769 ms | 149 - 149 MB | NPU
69
+ | MiniLM-v2 | ONNX | float | Snapdragon® 8 Gen 3 Mobile | 1.231 ms | 0 - 93 MB | NPU
70
+ | MiniLM-v2 | ONNX | float | Snapdragon® 8 Gen 1 Mobile | 2.535 ms | 0 - 98 MB | NPU
71
+ | MiniLM-v2 | ONNX | float | Qualcomm® QCS8550 (Proxy) | 1.772 ms | 0 - 3 MB | NPU
72
+ | MiniLM-v2 | ONNX | float | Qualcomm® QCS8450 | 2.535 ms | 0 - 98 MB | NPU
73
+ | MiniLM-v2 | ONNX | float | Snapdragon® 8 Elite Mobile | 0.866 ms | 0 - 62 MB | NPU
74
+ | MiniLM-v2 | ONNX | float | Snapdragon® 8 Elite Gen 5 Mobile | 0.698 ms | 0 - 58 MB | NPU
75
+ | MiniLM-v2 | ONNX | float | Qualcomm® QCS9075 | 2.964 ms | 0 - 45 MB | NPU
76
+ | MiniLM-v2 | ONNX | float | Qualcomm® QCS8750 | 0.866 ms | 0 - 62 MB | NPU
77
+ | MiniLM-v2 | ONNX | float | Qualcomm® QCS7181 | 1.769 ms | 149 - 149 MB | NPU
78
+ | MiniLM-v2 | ONNX | w8a8 | Snapdragon® X2 Elite | 0.669 ms | 213 - 213 MB | NPU
79
+ | MiniLM-v2 | ONNX | w8a8 | Snapdragon® X Elite | 1.788 ms | 149 - 149 MB | NPU
80
+ | MiniLM-v2 | ONNX | w8a8 | Snapdragon® 8 Gen 3 Mobile | 1.194 ms | 0 - 78 MB | NPU
81
+ | MiniLM-v2 | ONNX | w8a8 | Snapdragon® 8 Gen 1 Mobile | 1.98 ms | 0 - 78 MB | NPU
82
+ | MiniLM-v2 | ONNX | w8a8 | Qualcomm® QCS6490 | 4.839 ms | 0 - 45 MB | NPU
83
+ | MiniLM-v2 | ONNX | w8a8 | Qualcomm® QCS8550 (Proxy) | 1.716 ms | 0 - 27 MB | NPU
84
+ | MiniLM-v2 | ONNX | w8a8 | Qualcomm® QCS8450 | 1.98 ms | 0 - 78 MB | NPU
85
+ | MiniLM-v2 | ONNX | w8a8 | Snapdragon® 8 Elite Gen 5 Mobile | 0.664 ms | 0 - 56 MB | NPU
86
+ | MiniLM-v2 | ONNX | w8a8 | Snapdragon® 7 Gen 4 Mobile | 1.943 ms | 0 - 51 MB | NPU
87
+ | MiniLM-v2 | ONNX | w8a8 | Qualcomm® QCM6690 | 6.704 ms | 0 - 57 MB | NPU
88
+ | MiniLM-v2 | ONNX | w8a8 | Qualcomm® QCS9075 | 1.871 ms | 0 - 45 MB | NPU
89
+ | MiniLM-v2 | ONNX | w8a8 | Snapdragon® 8 Elite Mobile | 0.84 ms | 0 - 56 MB | NPU
90
+ | MiniLM-v2 | ONNX | w8a8 | Qualcomm® QCS7790 | 1.943 ms | 0 - 51 MB | NPU
91
+ | MiniLM-v2 | ONNX | w8a8 | Qualcomm® QCS8750 | 0.84 ms | 0 - 56 MB | NPU
92
+ | MiniLM-v2 | ONNX | w8a8 | Qualcomm® QCS7181 | 1.788 ms | 149 - 149 MB | NPU
93
+ | MiniLM-v2 | QNN_DLC | float | Snapdragon® X2 Elite | 0.682 ms | 1 - 1 MB | NPU
94
+ | MiniLM-v2 | QNN_DLC | float | Snapdragon® X Elite | 1.302 ms | 1 - 1 MB | NPU
95
+ | MiniLM-v2 | QNN_DLC | float | Snapdragon® 8 Gen 3 Mobile | 0.795 ms | 0 - 101 MB | NPU
96
+ | MiniLM-v2 | QNN_DLC | float | Snapdragon® 8 Gen 1 Mobile | 1.886 ms | 0 - 100 MB | NPU
97
+ | MiniLM-v2 | QNN_DLC | float | Qualcomm® QCS8275 | 3.561 ms | 0 - 58 MB | NPU
98
+ | MiniLM-v2 | QNN_DLC | float | Qualcomm® QCS8550 (Proxy) | 1.125 ms | 0 - 2 MB | NPU
99
+ | MiniLM-v2 | QNN_DLC | float | Qualcomm® QCS8450 | 1.886 ms | 0 - 100 MB | NPU
100
+ | MiniLM-v2 | QNN_DLC | float | Snapdragon® 8 Elite Mobile | 0.567 ms | 0 - 60 MB | NPU
101
+ | MiniLM-v2 | QNN_DLC | float | Qualcomm® SA8295P | 2.301 ms | 0 - 56 MB | NPU
102
+ | MiniLM-v2 | QNN_DLC | float | Snapdragon® 8 Elite Gen 5 Mobile | 0.461 ms | 0 - 60 MB | NPU
103
+ | MiniLM-v2 | QNN_DLC | float | Qualcomm® SA7255P | 3.561 ms | 0 - 58 MB | NPU
104
+ | MiniLM-v2 | QNN_DLC | float | Qualcomm® QCS9075 | 1.451 ms | 0 - 2 MB | NPU
105
+ | MiniLM-v2 | QNN_DLC | float | Qualcomm® QCS8750 | 0.567 ms | 0 - 60 MB | NPU
106
+ | MiniLM-v2 | QNN_DLC | float | Qualcomm® QCS7181 | 1.302 ms | 1 - 1 MB | NPU
107
+ | MiniLM-v2 | QNN_DLC | w8a8 | Snapdragon® X2 Elite | 0.819 ms | 1 - 1 MB | NPU
108
+ | MiniLM-v2 | QNN_DLC | w8a8 | Snapdragon® X Elite | 1.912 ms | 1 - 1 MB | NPU
109
+ | MiniLM-v2 | QNN_DLC | w8a8 | Snapdragon® 8 Gen 3 Mobile | 1.206 ms | 0 - 69 MB | NPU
110
+ | MiniLM-v2 | QNN_DLC | w8a8 | Snapdragon® 8 Gen 1 Mobile | 1.952 ms | 0 - 72 MB | NPU
111
+ | MiniLM-v2 | QNN_DLC | w8a8 | Qualcomm® QCS6490 | 3.699 ms | 0 - 2 MB | NPU
112
+ | MiniLM-v2 | QNN_DLC | w8a8 | Qualcomm® QCS8275 | 3.456 ms | 0 - 48 MB | NPU
113
+ | MiniLM-v2 | QNN_DLC | w8a8 | Qualcomm® QCS8550 (Proxy) | 1.715 ms | 0 - 51 MB | NPU
114
+ | MiniLM-v2 | QNN_DLC | w8a8 | Qualcomm® QCS8450 | 1.952 ms | 0 - 72 MB | NPU
115
+ | MiniLM-v2 | QNN_DLC | w8a8 | Snapdragon® 8 Elite Gen 5 Mobile | 0.681 ms | 0 - 54 MB | NPU
116
+ | MiniLM-v2 | QNN_DLC | w8a8 | Snapdragon® 7 Gen 4 Mobile | 1.694 ms | 0 - 47 MB | NPU
117
+ | MiniLM-v2 | QNN_DLC | w8a8 | Qualcomm® QCM6690 | 6.088 ms | 2 - 52 MB | NPU
118
+ | MiniLM-v2 | QNN_DLC | w8a8 | Qualcomm® QCS9075 | 1.862 ms | 0 - 2 MB | NPU
119
+ | MiniLM-v2 | QNN_DLC | w8a8 | Qualcomm® SA7255P | 3.456 ms | 0 - 48 MB | NPU
120
+ | MiniLM-v2 | QNN_DLC | w8a8 | Snapdragon® 8 Elite Mobile | 0.839 ms | 0 - 52 MB | NPU
121
+ | MiniLM-v2 | QNN_DLC | w8a8 | Qualcomm® SA8295P | 2.372 ms | 0 - 46 MB | NPU
122
+ | MiniLM-v2 | QNN_DLC | w8a8 | Qualcomm® QCS7790 | 1.694 ms | 0 - 47 MB | NPU
123
+ | MiniLM-v2 | QNN_DLC | w8a8 | Qualcomm® QCS8750 | 0.839 ms | 0 - 52 MB | NPU
124
+ | MiniLM-v2 | QNN_DLC | w8a8 | Qualcomm® QCS7181 | 1.912 ms | 1 - 1 MB | NPU
125
+ | MiniLM-v2 | TFLITE | float | Snapdragon® 8 Gen 3 Mobile | 0.832 ms | 0 - 102 MB | NPU
126
+ | MiniLM-v2 | TFLITE | float | Snapdragon® 8 Gen 1 Mobile | 1.932 ms | 0 - 102 MB | NPU
127
+ | MiniLM-v2 | TFLITE | float | Qualcomm® QCS8275 | 3.674 ms | 0 - 63 MB | NPU
128
+ | MiniLM-v2 | TFLITE | float | Qualcomm® QCS8550 (Proxy) | 1.127 ms | 0 - 2 MB | NPU
129
+ | MiniLM-v2 | TFLITE | float | Qualcomm® SA8775P | 12.459 ms | 0 - 28 MB | GPU
130
+ | MiniLM-v2 | TFLITE | float | Qualcomm® SA8650P | 12.459 ms | 0 - 28 MB | GPU
131
+ | MiniLM-v2 | TFLITE | float | Qualcomm® SA8255P | 12.459 ms | 0 - 28 MB | GPU
132
+ | MiniLM-v2 | TFLITE | float | Qualcomm® QCS8450 | 1.932 ms | 0 - 102 MB | NPU
133
+ | MiniLM-v2 | TFLITE | float | Snapdragon® 8 Elite Mobile | 0.593 ms | 0 - 65 MB | NPU
134
+ | MiniLM-v2 | TFLITE | float | Qualcomm® SA8295P | 2.366 ms | 0 - 56 MB | NPU
135
+ | MiniLM-v2 | TFLITE | float | Snapdragon® 8 Elite Gen 5 Mobile | 0.482 ms | 0 - 58 MB | NPU
136
+ | MiniLM-v2 | TFLITE | float | Qualcomm® SA7255P | 3.674 ms | 0 - 63 MB | NPU
137
+ | MiniLM-v2 | TFLITE | float | Qualcomm® QCS9075 | 1.501 ms | 0 - 45 MB | NPU
138
+ | MiniLM-v2 | TFLITE | float | Qualcomm® QCS8750 | 0.593 ms | 0 - 65 MB | NPU
139
+ | MiniLM-v2 | TFLITE | w8a8 | Snapdragon® 8 Gen 3 Mobile | 1.672 ms | 0 - 83 MB | NPU
140
+ | MiniLM-v2 | TFLITE | w8a8 | Snapdragon® 8 Gen 1 Mobile | 2.619 ms | 0 - 80 MB | NPU
141
+ | MiniLM-v2 | TFLITE | w8a8 | Qualcomm® QCS6490 | 13.129 ms | 0 - 31 MB | NPU
142
+ | MiniLM-v2 | TFLITE | w8a8 | Qualcomm® QCS8275 | 4.618 ms | 0 - 55 MB | NPU
143
+ | MiniLM-v2 | TFLITE | w8a8 | Qualcomm® QCS8550 (Proxy) | 2.366 ms | 0 - 3 MB | NPU
144
+ | MiniLM-v2 | TFLITE | w8a8 | Qualcomm® SA8775P | 12.541 ms | 0 - 30 MB | GPU
145
+ | MiniLM-v2 | TFLITE | w8a8 | Qualcomm® SA8650P | 12.541 ms | 0 - 30 MB | GPU
146
+ | MiniLM-v2 | TFLITE | w8a8 | Qualcomm® SA8255P | 12.541 ms | 0 - 30 MB | GPU
147
+ | MiniLM-v2 | TFLITE | w8a8 | Qualcomm® QCS8450 | 2.619 ms | 0 - 80 MB | NPU
148
+ | MiniLM-v2 | TFLITE | w8a8 | Snapdragon® 8 Elite Gen 5 Mobile | 0.947 ms | 0 - 54 MB | NPU
149
+ | MiniLM-v2 | TFLITE | w8a8 | Snapdragon® 7 Gen 4 Mobile | 5.123 ms | 0 - 35 MB | NPU
150
+ | MiniLM-v2 | TFLITE | w8a8 | Qualcomm® QCM6690 | 12.046 ms | 0 - 35 MB | NPU
151
+ | MiniLM-v2 | TFLITE | w8a8 | Qualcomm® QCS9075 | 2.498 ms | 0 - 24 MB | NPU
152
+ | MiniLM-v2 | TFLITE | w8a8 | Qualcomm® SA7255P | 4.618 ms | 0 - 55 MB | NPU
153
+ | MiniLM-v2 | TFLITE | w8a8 | Snapdragon® 8 Elite Mobile | 1.134 ms | 0 - 62 MB | NPU
154
+ | MiniLM-v2 | TFLITE | w8a8 | Qualcomm® SA8295P | 3.202 ms | 0 - 48 MB | NPU
155
+ | MiniLM-v2 | TFLITE | w8a8 | Qualcomm® QCS7790 | 5.123 ms | 0 - 35 MB | NPU
156
+ | MiniLM-v2 | TFLITE | w8a8 | Qualcomm® QCS8750 | 1.134 ms | 0 - 62 MB | NPU
157
+
158
+ ## License
159
+ * The license for the original implementation of MiniLM-v2 can be found
160
+ [here](https://github.com/UKPLab/sentence-transformers/blob/master/LICENSE).
161
+
162
+ ## References
163
+ * [Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks](https://arxiv.org/abs/1908.10084)
164
+ * [Source Model Implementation](https://github.com/UKPLab/sentence-transformers)
165
+
166
+ ## Community
167
+ * Join [our AI Hub Slack community](https://aihub.qualcomm.com/community/slack) to collaborate, post questions and learn more about on-device AI.
168
+ * For questions or feedback please [reach out to us](mailto:ai-hub-support@qti.qualcomm.com).
release_assets.json ADDED
@@ -0,0 +1,53 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "version": "0.57.0",
3
+ "precisions": {
4
+ "w8a8": {
5
+ "universal_assets": {
6
+ "tflite": {
7
+ "tool_versions": {
8
+ "qairt": "2.45.0.260326154327",
9
+ "litert": "1.4.4"
10
+ },
11
+ "download_url": "https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/minilm_v2/releases/v0.57.0/minilm_v2-tflite-w8a8.zip"
12
+ },
13
+ "qnn_dlc": {
14
+ "tool_versions": {
15
+ "qairt": "2.45.0.260326154327"
16
+ },
17
+ "download_url": "https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/minilm_v2/releases/v0.57.0/minilm_v2-qnn_dlc-w8a8.zip"
18
+ },
19
+ "onnx": {
20
+ "tool_versions": {
21
+ "qairt": "2.45.0.260326154327",
22
+ "onnx_runtime": "1.25.0"
23
+ },
24
+ "download_url": "https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/minilm_v2/releases/v0.57.0/minilm_v2-onnx-w8a8.zip"
25
+ }
26
+ }
27
+ },
28
+ "float": {
29
+ "universal_assets": {
30
+ "tflite": {
31
+ "tool_versions": {
32
+ "qairt": "2.45.0.260326154327",
33
+ "litert": "1.4.4"
34
+ },
35
+ "download_url": "https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/minilm_v2/releases/v0.57.0/minilm_v2-tflite-float.zip"
36
+ },
37
+ "qnn_dlc": {
38
+ "tool_versions": {
39
+ "qairt": "2.45.0.260326154327"
40
+ },
41
+ "download_url": "https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/minilm_v2/releases/v0.57.0/minilm_v2-qnn_dlc-float.zip"
42
+ },
43
+ "onnx": {
44
+ "tool_versions": {
45
+ "qairt": "2.45.0.260326154327",
46
+ "onnx_runtime": "1.25.0"
47
+ },
48
+ "download_url": "https://qaihub-public-assets.s3.us-west-2.amazonaws.com/qai-hub-models/models/minilm_v2/releases/v0.57.0/minilm_v2-onnx-float.zip"
49
+ }
50
+ }
51
+ }
52
+ }
53
+ }